prints (not just completed), matching get_printer_total_hours behavior
- Add option to keep or delete archives when deleting a printer
- Custom maintenance types no longer auto-assign to all printers
- Add UI to manually assign/remove custom maintenance types per printer
- Add backend endpoints for assigning types to printers and removing items
- Exclude static/assets from large file pre-commit check
- Use extruder_info[0].temp for right nozzle instead of nozzle_temper field which incorrectly mirrors left nozzle
- Gate dual nozzle parsing with has_h2d_extruder_info to prevent single nozzle printers showing two temps
- Add nozzle_heating calculation for single nozzle printers
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Add getStatusDisplay() helper in frontend to format status nicely
- Add get_derived_status_name() in backend to compute status from:
- stg_cur calibration stages (bed leveling, vibration comp, etc.)
- Temperature data when stg_cur unavailable (heating heatbed/nozzle)
- Include stg_cur and stg_cur_name in WebSocket broadcasts
- Update broadcast status key to detect heating phase changes
Printer cards now show: "Heating heatbed", "Heating nozzle", "Preparing",
"Auto bed leveling", "Cleaning nozzle tip", and 60+ other stage names
instead of just "Printing" or "Idle".
Highlights
This release focuses on stability improvements and bug fixes, with major fixes for browser freezing and UI responsiveness issues.
Bug Fixes
- Browser freeze on print completion - Fixed a critical issue where the browser would freeze when a print completed with the camera stream open. The fix uses buffered camera frames instead of spawning duplicate ffmpeg processes.
- Printer status "timelapse" effect - Fixed an issue where navigating to the printer page after a print completed showed metrics animating slowly from mid-print values to the final state. Printer status messages now bypass the throttled queue for immediate updates.
- Timelapse auto-download - Complete rewrite with retry mechanism and support for multiple storage paths.
- Timelapse detection for H2D - Fixed detection using the correct ipcam.timelapse field instead of xcam.timelapse.
- Reprint from archive - Fixed bug where the print button sent the slicer source file instead of the sliced gcode.
- Import shadowing bugs - Fixed ArchiveService import shadowing causing "cannot access local variable" errors.
New Features
- Failure reason detection - Automatically detects failure reasons from HMS errors (filament runout, layer shift, clogged nozzle)
- Hide failed prints filter - Toggle to hide failed/aborted prints with localStorage persistence
- Docker test suite - Comprehensive tests for build, backend, frontend, and integration
- Pre-commit hooks - Ruff linter and formatter for code quality
- Code quality tests - Static analysis to catch import shadowing bugs automatically
Changes
- Timelapse viewer default playback speed changed from 0.5x to 2x
- Archive badges now show "cancelled" for aborted prints
- WebSocket throttle reduced to 100ms for smoother updates
- Added ffmpeg to Docker image
Upgrading
cd bambuddy && git pull && docker compose up -d --build
Or use the auto-update feature in Settings.
When navigating to the printer page after a print completed, status
metrics would animate slowly from mid-print values to final state.
Root cause: The message queue processed messages one at a time with 16ms
delays. If 50+ status updates queued during a print, they took 800ms+ to
process, causing multiple throttle windows and the visual "timelapse".
Solution: Handle printer_status messages directly without queueing.
They're already throttled (100ms) so don't need queue rate limiting.
Other message types still use the queue.
Root cause: When a print completed with the camera stream popup open,
spawning a second ffmpeg process for finish photo capture caused a
conflict that froze the browser tab and video window.
Solution: Buffer the last frame from active camera streams. When
capturing finish photo, use the buffered frame if a stream is active
instead of spawning a new ffmpeg process.
Backend changes:
- camera.py: Added _last_frames buffer and get_buffered_frame() helper
- main.py: Photo capture uses buffered frame when stream is active
- main.py: Moved slow operations to background tasks (energy calc,
photo capture, smart plug, notifications, maintenance)
- archive.py: Fixed sync file write blocking event loop (asyncio.to_thread)
- printers.py: Added debug endpoint to simulate print completion
Frontend changes:
- useWebSocket.ts: Throttled printer status updates (100ms) to prevent
UI overload from rapid WebSocket messages
- useWebSocket.ts: Debounced archive invalidations (3s) to prevent
cascade of re-renders on print completion
- useWebSocket.test.ts: Updated tests for throttled/debounced handlers
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Backend (backend/app/api/routes/archives.py):
- Added filament_colors extraction from slice_info.config in 3MF files
- Maps filament IDs (1-based) to tool numbers (0-based)
Frontend (frontend/src/components/GcodeViewer.tsx):
1. Multi-color support: Pass extrusionColor as array of CSS color strings for multi-color prints
2. Bambu T-command filtering: Filter out special Bambu commands (T255, T1000, T1001, T65535, T65279) that corrupt tool state
3. Initial tool: Prepend T0 to ensure initial tool is set correctly
4. Disable gradient: Set disableGradient: true for multi-color to preserve actual filament colors (the gradient was turning black into gray/white)
5. Disable topLayerColor: Don't set topLayerColor for multi-color (it overrides per-tool colors)
Key fixes for Bambu G-code compatibility:
- Bambu uses special T commands (T1000, T65535, etc.) that aren't real tool changes
- The gcode-preview library's brightness gradient was modifying colors based on layer index
Commit message suggestion:
Add multi-color filament support to G-code viewer
- Extract filament colors from 3MF slice_info.config
- Pass color array to gcode-preview for tool-based coloring
- Filter Bambu special T commands (T1000, T65535, etc.)
- Disable gradient to preserve actual filament colors
When a print completes, multiple WebSocket messages are sent rapidly
(print_complete, archive_updated, etc.). Each was immediately invalidating
the archives query, causing a cascade of rapid re-renders that crashed
the browser tab.
Added debounced query invalidation to useWebSocket.ts:
- Multiple invalidation requests within 100ms are coalesced into one
- Uses requestAnimationFrame to avoid blocking the main thread
- Properly cleans up timeouts on unmount
- Cache connected state in ref to prevent flicker when status
briefly becomes undefined during query refetches
- Use cached value for shouldHide calculation instead of raw status
- Replace manual error database with 853 codes from ha-bambulab
(Source: https://github.com/greghesp/ha-bambulab)
- Filter out unknown HMS error codes from display
(Printers sometimes send duplicate/undocumented codes for same error)
- Update HMS badge count to only show known errors
- Export filterKnownHMSErrors helper for reuse across components
- Add ~80 HMS error code descriptions covering:
- 0300 series: temperature, fan, motor, force sensor, build plate errors
- 0500 series: storage, network, module communication, camera errors
- 0700 series: AMS communication, filament, hotend matching errors
- 0C00 series: toolhead camera, lidar, AI detection errors
- 1200 series: external spool errors
- Fix 0500_0400_0002_0043 description (toolhead camera dirty, not file system)
- Always link to HMS home page (wiki.bambulab.com/en/hms/home) instead of
specific error pages that may not exist
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- The failure analysis service was only querying status == "failed"
- Changed all queries to include both failed AND aborted: status.in_(["failed", "aborted"])
- File: backend/app/services/failure_analysis.py
Issue 2: Cannot clear failed flag in edit mode
- Added status field to ArchiveUpdate schema (backend)
- Added status to API client type (frontend)
- Added status dropdown to EditArchiveModal with options: Completed, Failed, Cancelled, Printing
- When status changes to "completed", failure_reason is automatically cleared
- File changes:
- backend/app/schemas/archive.py - added status field
- frontend/src/api/client.ts - added status to updateArchive type
- frontend/src/components/EditArchiveModal.tsx - added status dropdown
- Initialize i18n in test setup to suppress 'NO_I18NEXT_INSTANCE' warning
- Update AMSHistoryModal tests to use actual translated text
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
- Added DEBIAN_FRONTEND=noninteractive to suppress debconf warnings
- Added --root-user-action=ignore to pip install
- Added ffmpeg to Docker image
- Increased chunkSizeWarningLimit to 3000 for Vite build
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>
Test Summary:
- Build tests: 3 passed (image build, backend imports, static files)
- Backend unit tests: 378 passed (9 docker tests excluded)
- Frontend unit tests: 137 passed
- Integration tests: 9 passed (health, API endpoints, persistence, WebSocket)
Changes made to fix the Docker test suite:
1. Added curl to the production Dockerfile for integration tests
2. Removed deprecated version attribute from docker-compose.test.yml
3. Added --pull flag to all build commands to ensure fresh images
4. Added explicit build step before starting integration container
5. Fixed WebSocket test to accept 200 as a valid response
6. Excluded docker-marked tests from backend unit test runs (-m "not docker")
- H2D sends timelapse status in ipcam.timelapse field, not xcam.timelapse
- Added parsing for timelapse from ipcam data
- Added test to prevent regression
2. Systemd service updated - /etc/systemd/system/bambuddy.service
- Now runs as claude user instead of root
- Added ExecStartPre commands with + prefix (runs as root) to:
- Create logs and archive directories if missing
- Fix ownership to claude:claude before service starts
The service restarted successfully with the permission fixes running at startup. On the next print with timelapse enabled, both the snapshot and timelapse should be captured automatically.
1. bambu_mqtt.py: Added hms_errors to the completion callback data, which includes the HMS error codes when a print fails
2. archive.py: Updated update_archive_status() to accept an optional failure_reason parameter
3. main.py: Added auto-detection of failure reasons:
- status == "aborted" → failure_reason = "User cancelled"
- status == "failed" with HMS errors → maps module codes to reasons:
- Module 0x07 (Filament) → "Filament runout"
- Module 0x0C (Motion Controller) → "Layer shift"
- Module 0x05 (Nozzle) → "Clogged nozzle"
Frontend changes:
1. ArchivesPage.tsx:
- Badge now shows "cancelled" for aborted prints, "failed" for failed prints
- "Failed Prints" collection and "Hide Failed" filter now include both failed and aborted statuses
2. EditArchiveModal.tsx: Failure reason field now shows for both failed and aborted prints
New tests added:
- test_hms_errors_included_in_failed_completion_callback
- test_aborted_status_when_cancelled
The HMS error module mapping is basic and can be expanded as you observe more failure types. The actual Bambu HMS error codes are documented in their wiki - we can add
more mappings as needed.
- frontend/src/components/TimelapseViewer.tsx:17 - Changed from 0.5 to 2
2. Archive page: hide failed prints filter with persistence
- Added hideFailed state initialized from localStorage.getItem('archiveHideFailed')
- Added localStorage persistence via useEffect
- Added "Hide Failed" toggle button (red styling when active) next to the Favorites button
- Filter is automatically bypassed when viewing the "Failed Prints" collection
- "Reset" button also clears this filter
The filter state persists across page refreshes and browser sessions via localStorage.
Problem: xcam data was parsed BEFORE _was_running was set to True, so timelapse detection failed when both arrived in the same message.
Fix: Added timelapse detection at two critical points:
# Lines 1375-1381: When entering RUNNING state
if self.state.state == "RUNNING" and current_file:
if not self._was_running:
if self.state.timelapse:
self._timelapse_during_print = True
logger.info(f"[{self.serial_number}] Timelapse detected when entering RUNNING state")
# Lines 1391-1399: When print start detected
if self.state.timelapse:
self._timelapse_during_print = True
logger.info(f"[{self.serial_number}] Timelapse detected at print start")
2. Callback Error Handling (main.py + printer_manager.py)
Problem: The on_print_complete callback was hanging silently - no logs after "Found archive".
Fixes:
- Added try/except wrapper around "Update archive status" section (line 749-767)
- Added [CALLBACK] log messages at start and end of callback
- Added exception handler to _schedule_async in printer_manager.py to log any exceptions from async callbacks
3. New Integration Tests (test_bambu_mqtt.py)
Added TestRealisticMessageFlow class with 5 new tests that simulate real MQTT message sequences:
1. test_timelapse_detected_at_print_start_in_same_message - The critical race condition test
2. test_timelapse_not_detected_when_disabled
3. test_timelapse_detected_when_enabled_after_print_start
4. test_print_complete_includes_timelapse_flag_full_flow - Full lifecycle test
5. test_print_failed_includes_timelapse_flag
These tests process messages through _process_message() with proper {"print": {...}} wrapping, matching real MQTT flow.
Files Modified:
- backend/app/services/bambu_mqtt.py - Timelapse race condition fix
- backend/app/main.py - Error handling + logging for on_print_complete
- backend/app/services/printer_manager.py - Exception capture in _schedule_async
- backend/tests/unit/services/test_bambu_mqtt.py - New integration tests
Next Steps
You should restart the application to apply these changes. The next print will show:
- [CALLBACK] on_print_complete started for printer X
- [ARCHIVE] Updating archive X status...
- [ENERGY], [PHOTO], [AUTO-OFF] logs (if those features trigger)
- [CALLBACK] on_print_complete finished for printer X, archive X
Changes to ToastContext.tsx:
- Added 'loading' toast type with a spinning Loader2 icon
- Added showPersistentToast(id, message, type) function for toasts that don't auto-dismiss
- Exposed dismissToast(id) function to allow programmatic dismissal
- Added green-themed styling for loading toasts
Changes to SettingsPage.tsx:
- When exporting a backup that includes archives, shows a persistent "Preparing backup..." toast with spinner
- The toast is automatically dismissed when the download starts or if an error occurs
- For non-archive backups (which are fast), no loading toast is shown
The user will now see clear feedback when creating a backup with print archives - a toast with a spinner appears immediately after clicking export and stays visible
until the download dialog appears.
Expanded the useWebSocket.test.ts from minimal coverage to 20 comprehensive tests covering:
WebSocket Mock Tests (6 tests)
- Creates WebSocket with correct URL
- Starts in CONNECTING state
- Transitions to OPEN state
- Can receive messages
- Can close connection
- Tracks all instances
Hook Connection Tests (2 tests)
- Connects to WebSocket on mount
- Reports connected state when WebSocket opens
Message Handling Tests (9 tests)
- Updates printer status in query cache on printer_status message
- Preserves wifi_signal when new value is null
- Invalidates archives on print_complete message
- Invalidates archives on archive_created message
- Invalidates archives on archive_updated message (new handler for timelapse auto-assignment)
- Ignores pong messages without error
- Handles malformed JSON gracefully
- Handles unknown message types gracefully
sendMessage Tests (2 tests)
- Sends JSON message when connected
- Does not send when disconnected
Reconnection Tests (2 tests)
- Reconnects after connection closes
- Cleans up on unmount
Key Fixes
- Fixed MSW (Mock Service Worker) conflict by:
a. Adding addEventListener/removeEventListener to MockWebSocket class
b. Updating MSW setup to bypass WebSocket requests
c. Properly managing WebSocket mock lifecycle in each test
Test Results:
- Frontend: 137 tests passed (9 test files)
- Backend: 346 tests passed
Problem: At print start, the MQTT mc_remaining_time was often 0 or unavailable because the printer hadn't fully processed the gcode yet.
Solution: Modified the notification system to prioritize the print_time_seconds from the 3MF file parsing (which is always available after archive creation) over the
MQTT remaining time.
Files changed:
- backend/app/services/notification_service.py:
- Added optional archive_data parameter to on_print_start() method
- Added priority logic: archive data > MQTT remaining_time > raw_data mc_remaining_time
- backend/app/main.py:
- Created _send_print_start_notification() helper function
- Moved notification sending to AFTER archive creation (when print_time_seconds is available)
- Added fallback notification sending at early return points (auto_archive disabled, file not found, etc.)
Issue 2: Timelapse Auto-Assignment
Problem: Timelapses couldn't be auto-assigned reliably because printer dates could be off in LAN-only mode, making timestamp matching unreliable.
Solution: Track whether timelapse recording was active during the print via MQTT, and if so, auto-fetch the most recent timelapse file on print completion.
Files changed:
- backend/app/services/bambu_mqtt.py:
- Added _timelapse_during_print instance variable
- Track when timelapse is active during a print (in both data and xcam parsing)
- Reset the flag on new print start
- Include timelapse_was_active in the print complete callback data
- backend/app/main.py:
- Added timelapse auto-scan logic in on_print_complete()
- If timelapse_was_active is True and status is "completed":
- Wait 5 seconds for timelapse to finalize
- Scan printer's timelapse directory
- Find most recent file by mtime
- Verify it was modified within 30 minutes of print completion
- Download and attach to archive
- Send WebSocket update
- Header navigation link
- Installation guide link
- Documentation section with all 5 doc links
- Footer documentation link
2. Screenshots updated with new images:
- Replaced old maintenance screenshots with maintenance-1.png and maintenance-2.png
- Replaced old profile screenshots with cloud_profiles-1.png, cloud_profiles-2.png, k_profiles-1.png, k_profiles-2.png
- Replaced old settings screenshots with settings_general.png, settings_smart_plugs.png, settings_notifications.png, settings_api_keys.png
- Removed old unused screenshot files from docs/screenshots/
maziggy